0:00 Good morning, everybody. I'm Doug Straton. I lead marketing here at Bazaarvoice. Just a little background. I'm a former client, running digital and digital transformation for a couple big CPGs. 0:11 Wanted to spend, some time with you today and and talk a little bit about how, people find products today. So the overall shopping journey still looks really, really familiar. 0:22 People research, compare, and decide, but where all that information is surface surfaces has changed. And more and more of discovery now runs through AI in one form of of or another. 0:34 When I say discovery, I mean the point where shopper first encounters your product and starts to form confidence in and around it, whether that happens to be on a PDP, through a search result, or now in an AI summary or shopping agent. 0:47 And it's the same job, quite frankly, that the retail shelf does, making products visible and comparable. It's just that the modalities are moving a bit around. 0:55 And what I wanna do today is walk through what that shift means for you, where reviews and UGC really start to matter, and how we think about staying visible, influencing confident shopper decisions, and ultimately helping them choose your product in this new version of the shelf. 1:11 So we'll get into it right now. And we'll start with a question that I get asked frequently. How do we make sure our products still get be get seen by shoppers and by AI? 1:21 And the short answer is discipline, content discipline, and data discipline. So authentic and trusted content, like strong ratings and news and a steady flow of UGC are now part of your core infrastructure. 1:33 It's as fundamental as your core creative, your product data, and the content supply chain that holds those things together. 1:42 It's really how AI learns to see you. And it's important because AI is reading the same content your shoppers read, and it uses that content to decide which brands it can recommend with confidence. 1:53 So your data and your UGC together form your new foundation for visibility. So let's take a quick look at what's changed. 2:04 Roughly seventy percent of Google queries now come back with an AI overview. Organic traffic from those listings is down. 2:12 It can be as down as far as sixty percent in some cases depending how you're measuring it. And shoppers aren't scrolling through ten options anymore. They're seeing two, three, four, maybe five that an AI agent has already filtered, And that's essentially your new shelf. 2:27 So conversational search, minimal scroll, and and fewer clicks, if not zero clicks. And if you're not on that short list, you're not necessarily gonna end up in the cart. 2:37 But it's really not all bad news. The shoppers who use those AI summaries, what we call LLM shoppers, convert about four and a half times higher than organic search shoppers. 2:46 So the value of showing up is actually increasing even though the space to show up keeps shrinking. So the game's shifting a bit where it used to be you wanna be eye level on the shelf. 2:57 It then became let's make make sure that we're the page one of search results, and now it's ensuring that we get into the summary itself. And that's where the shopper's attention is gonna live, and this is where your AI strategy really needs to be built to influence. 3:13 So people, you know, typically can go right right to panic. And I think, you know, from a leadership perspective, I know in my former life when big technological shifts would happen, there's always the question about, like, what's going to happen, what are we doing about it, and how can we make the best of this situation. 3:30 And I'm here to tell you that that AI driven discovery does not equal doomsday. It's not the end of discovery. It's just an evolution. We've moved shelves before from physical shelves to search pages now to AI summaries and shopping agents. 3:44 And really, it's the same principle. If you understand how shoppers search, you win the placement. So your AI strategy should really be guided by your own shopper insights and how to improve AI shelf visibility within those insights. 3:58 So all those about shopper insight studies, consumer insight studies, all the things that you've been that you've been really leaning on for years and years to drive decisions about how you proactively manage your portfolio in a physical world and or a digital world still apply. 4:12 So let's take a second and just think about this for for a minute. The first thing you need to know is that AI is simple. It's just the next evolution of input and output. 4:22 You shouldn't be intimidated about this. It feels really complicated, but it it really is not. Information goes in and an answer comes out. 4:31 What decides the quality of that answer is what you feed into it. The structure, accuracy, and freshness of your content, whether it's brand creative or user generated content like reviews. 4:42 And when you think about an AI strategy, don't start with the tool or the prompt. Start with the inputs. What are you gonna give it to work with? How current, complete, and consistent is that data? 4:51 And when your product information is clean and your UGC is strong, AI will represent you and and surface you very well. 5:00 But when the inputs are thin or the data your data is inaccurate, it's gonna fill in blanks. It's gonna start guessing. And it's usually with somebody else's story, not necessarily yours. 5:11 So focusing on the inputs, that's where you can build in a bit of control over the outputs. Now the second thing is is that AI needs to learn. 5:21 For AI to give a good answer, it and to represent your brand accurately, it needs to learn from the right signals. And those signals essentially come from places that you already know, like SEO optimization, your branded content, the quality of the the text and copy that you're using to build your PDPs, catalog accuracy. 5:44 And the second thing you need to understand is that verified reviews in UGC are anchors for AI. And we're gonna get to get to a little bit more of the information behind that in a few minutes. 5:56 But it's really what centers your content and really drives recommendation for summaries as well as for agentic shopping. 6:05 The other thing to keep in mind is that regulations will continue to raise the bar for authenticity. So the regulations haven't quite caught up to what's going on in the marketplace, but over time, you can predict just like your branded content and just like your user generated content, that governments all over the world are gonna react and, and start to think about how do they protect consumers' interest in this new world. 6:30 So net net, you need a strategy about how you want to use the accessibility or allow and use the accessibility of your data and use your source of trust to train AI. 6:39 Now the third thing is AI wants to be trusted. And this is the people that really surprises people, but it's the it's the key that that connects all this together. 6:50 AI really just behaves a lot like shoppers. It wants confidence in what it recommends before it recommends anything. 6:58 And so the way you need to think about this is the the place where it's pulling in a lot of that confidence, quite frankly, is from ratings, particularly the coverage and review volume. 7:10 So your average rating, the number of reviews really strengthens the confidence of the AI model in picking or recommending or summarizing a specific product. 7:21 And so the ability to actually build up again that that that library of content behind your product catalog as it relates to user generated content is a key key component of scaling of authenticity, and then ultimately summaries, and agentic recommendations. 7:40 So the the key thing is is is this is something that actually we've been talking about for years, twenty years, in fact, in terms of ratings and reviews and broader UGC. 7:48 AI ultimately acts like humans in trusting authentic user generated content. Humans look at the same things, volume, average rating to start to formulate an opinion and AI is essentially picking up on the same behavior when it's making decisions and making recommendations. 8:06 Now reviews of the new currency. So, an article from August fourth, from a joint study between Yale and Columbia and since quoted by, Carey Masters has really brought home the importance of user generated content and specifically ratings or re reviews in the context of AI summaries and AgenTeq recommendations. 8:30 So what they did is they looked at how AI agents, Claude, GPT four dot one, Gemini, decide which products to recommend and what kind of pricing they'll stand behind or or or how they'll think about the recommendations. 8:42 And when review volume went up, basically, you're doubling the amount of feedback that was coming into the system, the models were still really comfortable recommending the product even if in fact the price was not as low. 8:55 Just think of it in terms of the more content you have, it's the same as like giving a discount essentially to a consumer because it's gonna be the product that AI trusts and then therefore recommends. 9:06 And that's really that's really remarkable. And average ratings just a just a point one or a onetenth of one percent increase in average ratings had similar impacts. 9:18 So it's pretty remarkable. It's causal. It's not correlated. It means that the systems are not only reading reviews, they're using this proof of value. And so the more credible structured UGC that you feed in, the more confidence these models are gonna have in suggesting your product and brand. 9:33 So the shift that we're talking about is really interesting because the way you need to think about this is that you you essentially have two consumers of confident or or of of content. 9:44 You've got humans consuming content, and that's one audience. And then there's machines consuming content, and that's the other audience. And that second audience is the audience that acts as somewhat of a gatekeeper. 9:57 So it's no longer just about that shopper making the decision. It's about the AI systems interpreting information on their behalf, and the quality of what they learn from you determines whether you're gonna be service recommended or ignored. 10:10 And so your brand and your storytelling and your creativity still really matter, but the foundation of discoverability today is really user generated content. 10:19 That's what gives AI a factual, human, authentic, trustworthy signal. It's how the model learns who you are, what you stand for, and whether it should treat your products as credible. 10:30 So inside of Bazaarvoice, we look at this through what we call AAA, accessible, authentic, abundant content. It's essentially an operational checklist that underpins everything we do. 10:41 And accessible means that your data and content are structured, connected, and machine readable across every surface, whether it's a retailer site, Google AI overviews, the broader ELLM ecosystems like ChatGPT, Perplexity. 10:54 And if your information is fragmented or inconsistent, those systems simply can't see you. So authentic is verified human experience. 11:03 That's a differentiator. In an AI first world, authenticity becomes a data quality issue. And we remove every detected instance of fraud from billions of reviews because if the input is incredible, the output can't be trusted by either the people or by the machines. 11:21 And our final a, abundant, is really is about scale and relevance. These systems learn through repetition and recency. They need a constant flow of credible content to build confidence in your brand. 11:32 A dozen reviews doesn't necessarily move the dial. So depth and freshness across your key products will, in fact, move that dial. And AAA is how you make your brand intelligible to AI. 11:44 So they can read you accurately, represent you fairly, and continue to recommend you. It's really the practical groundwork for visibility and trust in this new environment. 11:55 Now once you understand that framework, the question is is always where do I start? Because you can't really fix everything at once. 12:03 And the best way I think, in in my experience, to think about this is how I would think about any kind of big transformation is through the lens of my portfolio. 12:13 So, you know, if if when I think about this, I think about, like, what are the twenty percent of the products that are driving eighty percent of my sales and eighty percent of my profit? 12:21 Then I I'll I'll start there. And you focus, you know, your visibility efforts on that area first. 12:30 And that's where, again, improving data quality, review volume, and freshness is gonna deliver probably the fastest return fastest return. And from there, can think about innovation. 12:41 So innovation products, are your new launches. Yeah. You have to make sure that they get early authentic review volumes so that AI learns about them quickly and doesn't default to older items. 12:51 And, obviously, renovation, the new packages new packaging or restages of products and brands, super super important as you're moving through those processes that you start to work on those and connect them carefully to the past data that these models might have been reading so that they don't lose context or treat them as as as absolutely new products when in fact they have long histories. 13:15 So if you apply that discipline, protect your brand's discoverability and pricing power across the whole portfolio, You won't get blindsided by, I would say, you know, like, a little bit losses in visibility across the landscape as you're chasing the next big thing in discoverability. 13:34 So I I think this last piece is, super important. So, we've talked about the the the triple a framework, making sure that you've got, accessible, authentic, and abundant content that that that your data quality and integrity, the things that you've been working on anyway to to optimize for for digital over the past couple years just becomes that much more important. 13:57 And we've now talked about how you can approach it from a portfolio perspective. So focusing on your core, focusing on your innovation, and focusing on your renovation, and just use that as a as a process to gradually improve the visibility of your your total portfolio over time in a way that's really responsible. 14:15 But the the one thing that I think is really fundamentally important and to think about in a really deep way is think about the modality of AI. 14:24 The modality is you put a deep question in and you get a deep answer out. So question, answer. And as it turns out, just as you've been optimizing for keywords in the traditional search environment, optimizing for questions and optimizing for answers is what LLMs are really designed to do. 14:44 And if you think about optimizing through that lens of q and a, that can be really profound in terms of the way you approach your overall strategy as it relates to AI and discoverability in the age of AI. 15:00 So I'm happy to take some questions. I think they're already popping in here. First one, are there specific benchmarks or activity thresholds like a certain number of views, minimum star rating within a set time period that help AI systems recognize and recommend a product. 15:17 So at this point, I think there's there's that's a little bit of the the unknown. But what I would suspect that we're gonna see is the things that we've been seeing as we've, you know, mined for insights as it relates to human behavior at a categorical level. 15:34 So to answer your question, it's it's gonna it's gonna be different for every single category, and it might even be difficult different by by vertical. 15:43 So there'll be different thresholds. In speaking with the LLMs, we know that they're thinking about those things, but they haven't yet actually ascertained specifically how they're gonna use that to influence the models. 15:57 It's still very early days. So even in this sense that they are using, for example, ratings and reviews data, it's much more superficial than you might think. 16:07 And this is why it's so important for for for us to build this corpus of reviews and broader user generated content to marry up with your brand creative and your product information in a way that helps these models understand things. 16:22 So no no standard or benchmark just yet as it relates to volume or ratings, but in my view, it will be different by vertical. It'll also be different by category. 16:31 Another question. How can brands effectively gather more verified reviews and user generated content while making sure the content remains authentic and trustworthy for both shoppers and AI? 16:44 So I think, you know, all your your standbys here, you can there's there's sampling programs or, you know, value exchanges that you can make with consumers and ask them for reviews through through broader efforts. 16:58 It might be through, like, your CRM efforts or something like that. There's also the curation of content through social, something that we also do. 17:07 Those are great ways to think about how you would get the kind of content you need to build up that review corpus to drive AI. 17:16 And those are things you're probably already doing. So it kinda goes back to thinking about the things that you're already doing to optimize. They're still in play. There's not necessarily new tactics. 17:25 There's just you need to be even more diligent about the structure, and the integrity of the data that you're using to feed your broader digital environments because they're gonna get picked up. 17:37 Okay. Lots of questions coming here. Let's see here. Can you elaborate the weight AI models put on answers to questions? 17:47 Should we on the side of detailed keyword stuffed answers to please the a AI model? So, yeah, this is a really interesting question. I would say the practice on this is also evolving. 17:57 This is how new the space is. But I think there's gonna be the same sensitivity over time to things like question and answer stuffing that there is to, like, keyword stuffing. 18:09 They're gonna recognize that things are being that the systems are being gamed. So what I would what I would say here is you just have to naturally understand, again, I think your shopper behavioral insights and your consumer insights as they relate to how people interact with your brand, product, and as as probably a set especially the category. 18:33 And then ensure that you're understanding, whether through its social listening or maybe you have q and a as it as it relates to the functionality that you're either using through Bazaarvoice or some other providers, and understand what people are already asking in the retail environments and how you're already answering those questions because those can be deep in and insightful, I would say, minds of data that you could go into and to try to figure out what your strategy is gonna be. 18:59 But I wouldn't go I wouldn't go truly to stuffing, and I wouldn't I wouldn't, suppose at this point in time that we even have enough data to understand what stuffing looks like in this new environment. 19:08 But clearly, people will be thinking about, that type of a tactic. Let's see here. Lots of questions. Let's see here. 19:19 For brands that are newer to the market and competing against established players with much higher review volumes, how would you recommend we stay competitive without breaking the bank on reviews programs? 19:30 That's a super good question. I do think, in general, big brands, at least initially, will have some sort of an advantage because of the volume of content that they generate. 19:42 I think over time that as the models get better and they start they start thinking about the recency and and going back to what I was saying before in terms of different categorical behaviors, like, you know, is the is the recency for one category, like, a a review needs to be within six months versus a three months or whether it's four weeks or whatever it might be. 20:03 I think there'll be differences there. The LLMs are not necessarily capturing that data just yet and and they're doing a lot to to use that, but that will that will come over time. 20:12 So I think depending on your category, it it might be that recency of reviews becomes more important than overall volume. But I but I think in general, the things that you can think about if you're a newer brand is think about your overall product launch or introduction and the activities that you're already doing that will in fact also influence the LLM. 20:34 So the LLMs we do know are very sensitive in terms of picking up articles and citations from trusted sources and as well as using the product data. 20:47 So I think you you have to think about again that the comments that I made earlier in the session around SEO optimization, making sure you're brand creative, your product detail pages, get all of that right. 20:58 But then there's also a whole bevy of other activity that you have in the digital space that likely accrues in your favor. So it could be PR mentions or or something like that. 21:08 That'd be something that I would concentrate on. Alright. Let's see here. Read more. We need to optimize for the AI crawler bots for learning grand mentions, but also on demand web searches that AI performs answer a specific prompt. 21:21 How do you recommending ensuring UGC is accessible to LLMs for both scenarios? If I'm reading the question correctly, there there is there is two things that are going on. 21:30 There is and this this is a really smart question. So if I'm interpreting it correctly, there's an element of LLMs where they're going out and they're crawling for data and they're using that data to train their their models. 21:45 In other cases, they might have access to data that they can use in real time that might have, I would say, like, prepopulated summaries or something that's related to the individual product and can can be connected in real time to, a number of queries that a person or, in this case, a shopper may be throwing at the the LLM. 22:06 So I think there's a little bit of both. I think when I when I'm talking about authentic, accessible, and abundant content, that's making it broadly visible and discoverable by crawlers so they can pick those things up and start to learn from them and then understand when they should be recommending them. 22:23 I think the other element, I think, of what your question alludes to is still early days where the LLMs contextually, particularly the agentic shopping engines, will actually go and pull information in real time out of trusted sources like Bazaarvoice to answer certain questions that are being posed by the user. 22:44 So I'm not sure I'm answering the question, but I'm getting in that that the the overall space. Alright. Trying to see if there's a couple of other questions here. 22:55 Alright. I will go into just a quick note here on Summit. So we're learning a lot as we're going as we're going through this process just as you are and just as the LLM providers themselves are learning. 23:12 So everybody's on a big journey right now. It is moving incredibly fast. We're gonna continue to think about how we approach this in a way that's really beneficial to you, our clients. 23:23 So this last piece is really about how AI is, you know, actually interpreting what you put in the world. So, again, if you wanna understand more as we're learning more about these systems and and how they work and how they extract meaning and if you want the right answers to come back and to give those models the language that understands. 23:47 We're gonna be doing more of this kind of conversation in a deeper way during our Bazaarvoice Summit, which is January twenty first twenty twenty six. 23:56 This is our thirteenth edition remarkably. It's essentially a three hour virtual event. We'll talk in much more specificity about mastering AI first discovery, how to scale authentic content to power influence. 24:11 We we we talked about that infrastructure that you need to power AI first discovery, and then how to turn that discovery into revenue. 24:21 So there'll be specific playbooks in this case that show you the ROI of your content, how it drives purchases through the lens of AI. 24:29 So we'll have a number of guest speakers for that event. I will be attending that event as well. We'd love to have you there. Okay. I think I can go back quickly here for one more question. 24:40 It says, with the landscape shifting to video content, are LLMs now able to review UGC videos? So that is a super, smart question. 24:50 So what I would say here is right now the LLMs are still figuring out kind of the the basic shopping modalities. If you go into those a lot of those environments right now, the experience can range from, you know, slightly magical to somewhat frustrating. 25:06 So they still haven't completely sorted exactly how things are gonna work in these environments. But one thing that we do know is that while right now they're they're still focusing in on the basics that over time that, yes, they will want to understand and analyze both images and video, not just for content, but for context and insight as to how they should change the environment as you're actually inside of it and what they should be recommending. 25:36 So those things are important. As we've talked to them, they've all mentioned that these things are important. Obviously, we've been focusing in on the ability to pull in both images and videos for the past couple years as long as textual ratings and reviews. 25:52 So it's a big focus of ours. So it is a focus of the LLMs currently, but not necessarily something that's manifesting in the environment just yet. 26:03 But it will come, and it will come fairly quickly. Okay. With that, I think we'll call it a wrap unless there's any other questions. 26:12 Thank you so much for for spending some time with us. We really, really appreciate. I hope we see you at the summit on January twenty first twenty twenty six. 26:21 And if there's any questions or whatnot, reach out to your Bazaarvoice account directors or client success managers. 26:30 They're all being trained up just as fast as everybody is in the new world of AI discovery, and they'll be happy to help in any way, shape, or form bring your AI strategies to life. 26:40 With that, I will say thank you, and I will see you next time.
